| Summary: | To design a high resolution spectrum estimation module as part of a digital tracking array system, the theory and mathematical formulations of several high resolution spectrum estimation methods are presented. In the implementation of a spectrum estimation system, the received signal is first down-converted to baseband frequency using single channel or in-phase (I) and quad phase (Q) channel down-converter before it is digitized using an analog-to-digital (ADC) converter. Three distinct frequency estimation methods, namely multiple signal classification (MUSIC), estimation of signal parameters via rotational invariance techniques (ESPRIT) and multi-resolution spectrum sensing (MRSS), are simulated to detect the inherent frequencies of a test signal. The performances, such as estimation accuracy, frequency resolution, processing speed, observation time, and resilience to noise, are measured and evaluated. Comparing the simulation results, the MRSS out-performs the MUSIC and ESPRIT in terms of spectral resolution, estimation accuracy, and robustness to noise. Though the MRSS requires a higher observation time and processing time, the values remain significantly low at 13 us and 2.4 us, respectively for SNR equals to -10 dB. Hence, the MRSS is proposed as the frequency estimation algorithm in the digital tracking array to provide accurate, robust, and high resolution spectrum estimation.
|
|---|
